Random forest integration method improved through width neural network

A technology of random forest and integration method, applied in the field of random forest integration improved by wide neural network, can solve the problems of deep forest parallelization limitation, lack of rationality, etc., achieve high degree of automation, improve training efficiency, and strong parallelization ability Effect

Inactive Publication Date: 2018-11-09
CHINA UNIV OF MINING & TECH
View PDF0 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, under the multi-layer cascade structure of deep forest, the parallelization of deep forest will be greatly restricted; in addition, deep forest obtains the final output vector by solving the average value of each output vector at the output layer, which lacks certain rationality.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Random forest integration method improved through width neural network
  • Random forest integration method improved through width neural network
  • Random forest integration method improved through width neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0031] Specific embodiments of the present invention will be further described below in conjunction with accompanying drawings

[0032] A random forest integration method based on feature mapping layer and enhancement layer structure, characterized in that: it includes two parts of model design and model training;

[0033] For samples with a spatial position relationship between image or input data features, this method is used to generate the feature input vector of the feature mapping layer. For the input without spatial position relationship, the original input is directly used as the input feature vector of the feature mapping layer.

[0034] Such as figure 1 The overall flow chart of the model is shown, in which the input layer performs multi-granularity scanning on the data with spatial position relationship to obtain the input feature vector. However, for the original input that is not associated with the spatial position, it can be directly used as the input feature v...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a random forest integration method improved through a width neural network, and is suitable for the field of machine learning. The method mainly comprises two parts: model design and model training. The model design mainly comprises two parts: the design of a feature mapping layer and an enhancement layer, and the design of an output weight. A neural network node composed of a random forest and a complete random forest is designed so as to adaptively adjust the width of a model. A local weight is obtained through the mean accuracy of the nodes, and the output weight iscalculated, and finally a final output vector is solved. The method is high in automation degree, adaptively decides the size of the model through the training, is easy for theoretical analysis, is good in interpretability and is strong in parallelization capability.

Description

technical field [0001] The invention relates to a random forest ensemble method, in particular to an improved random forest ensemble method using a wide neural network used in the field of integrated machine learning. Background technique [0002] Machine learning is one of the hottest research fields at present. In recent years, with the continuous growth of data volume, the efficiency and accuracy of machine learning have attracted much attention. Ensemble learning has always been regarded as an effective method to improve the accuracy of the model, which is widely used in supervised learning and unsupervised learning. [0003] Recently, Zhi-Hua Zhou et al. proposed an integrated learning method based on random forest - deep forest (gcForest). Deep forest is a deep model other than convolutional neural network. Compared with deep convolutional neural network, it also has the ability of representation learning and has the following advantages: the model has fewer hyperpara...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04
CPCG06N3/045G06F18/24323G06F18/214
Inventor 刘鹏王学奎魏卉子尹良飞景江波叶帅仰彦妍
Owner CHINA UNIV OF MINING & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap